Write down the plural-key gotcha, and how to tell it happened #280

New Issue

Closed

opened 2026-09-04 19:28:16 +00:00 by jcoffey-dev · 0 comments

jcoffey-dev commented

2026-09-04 19:28:16 +00:00

Owner

There was no CLAUDE.md. This adds one, holding what has actually cost time in this repo rather than a style guide — CONTRIBUTING.md is that.

The plural key is the other form. plural() looks the entry up by forms.other, so a call site written as plural(n, { one: "Deleted {n} contact", other: "Deleted {n} contacts" }) is keyed on the plural. Keying the catalogue on the one form type-checks, builds, passes every test, and silently renders English in all nine languages. It cost a round trip on #278.

The part worth writing down is the signal, not the rule. There is exactly one: the "falling back to English" count from i18n-catalog-check. The percentage is actively misleading here — adding keys moves the denominator, so it sat at 98% whether or not the new strings were translated. The note says to compare the fallback count against main, and why not to read the percentage instead.

Two other things, both from this week:

A change touching user-visible strings is work in nine catalogues, and should be reported explicitly — including when the answer is "none added", so it is never left to be inferred.
Store tests do not exercise the component. The shift-click range bug in #278 is the standing example: the range was measured inside a setState updater, which React runs after the anchor ref has already moved, so it passed every store assertion and failed the moment the built app was driven.

Docs only — no code, no strings, no catalogue changes.

Merged 2026-09-04 as coffey-labs/ihasmail@fcbd8f6449

_{Rebuilt from: GH Archive, git history, session transcript.}

There was no `CLAUDE.md`. This adds one, holding what has actually cost time in this repo rather than a style guide — `CONTRIBUTING.md` is that. **The plural key is the `other` form.** `plural()` looks the entry up by `forms.other`, so a call site written as `plural(n, { one: "Deleted {n} contact", other: "Deleted {n} contacts" })` is keyed on the plural. Keying the catalogue on the `one` form type-checks, builds, passes every test, and silently renders English in all nine languages. It cost a round trip on #278. **The part worth writing down is the signal, not the rule.** There is exactly one: the "falling back to English" count from `i18n-catalog-check`. The percentage is actively misleading here — adding keys moves the denominator, so it sat at 98% whether or not the new strings were translated. The note says to compare the fallback count against `main`, and why not to read the percentage instead. Two other things, both from this week: - A change touching user-visible strings is work in nine catalogues, and should be reported explicitly — including when the answer is "none added", so it is never left to be inferred. - Store tests do not exercise the component. The shift-click range bug in #278 is the standing example: the range was measured inside a `setState` updater, which React runs after the anchor ref has already moved, so it passed every store assertion and failed the moment the built app was driven. Docs only — no code, no strings, no catalogue changes. **Merged** 2026-09-04 as coffey-labs/ihasmail@fcbd8f64491b <sub>Rebuilt from: GH Archive, git history, session transcript.</sub>

This repo is archived. You cannot comment on issues.

1 Participants

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: coffey-labs/ihasmail-github-archive#280